Image inpainting aims to fill missing regions of a damaged image with plausibly synthesized content. Existing methods for image inpainting either fill the missing regions by borrowing information from surrounding areas or generating semantically coherent content from region context. They often produce ambiguous or semantically incoherent content when the missing region is large or with complex structures. In this paper, we present an approach for image inpainting. The completion model based on our proposed algorithm contains one generator, one global discriminator, and one local discriminator. The generator is responsible for inpainting the missing area, the global discriminator aims evaluating whether the repair result has global consistency, and the local discriminator is responsible for identifying whether the repair area is correct. The architecture of the generator is an auto-encoder. We use the skip-connection in the generator to improve the prediction power of the model. Also, we use Wasserstein GAN loss to ensure the stability of training. Experiments on CelebA dataset and LFW dataset demonstrate that our proposed model can deal with large-scale missing pixels and generate realistic completion results.

/pdf/image-inpainting-based-on-generative-adversarial-networks-nixexiewps.pdf

Image Inpainting Based on Generative Adversarial Networks

The fast-growing amount of information on the Internet makes the research in automatic document summarization very urgent. It is an effective solution for information overload. Many approaches have been proposed based on different strategies, such as latent semantic analysis (LSA). However, LSA, when applied to document summarization, has some limitations which diminish its performance. In this work, we try to overcome these limitations by applying statistic and linear algebraic approaches combined with syntactic and semantic processing of text. First, the part of speech tagger is utilized to reduce the dimension of LSA. Then, the weight of the term in four adjacent sentences is added to the weighting schemes while calculating the input matrix to take into account the word order and the syntactic relations. In addition, a new LSA-based sentence selection algorithm is proposed, in which the term description is combined with sentence description for each topic which in turn makes the generated summary more informative and diverse. To ensure the effectiveness of the proposed LSA-based sentence selection algorithm, extensive experiment on Arabic and English are done. Four datasets are used to evaluate the new model, Linguistic Data Consortium (LDC) Arabic Newswire-a corpus, Essex Arabic Summaries Corpus (EASC), DUC2002, and Multilingual MSS 2015 dataset. Experimental results on the four datasets show the effectiveness of the proposed model on Arabic and English datasets. It performs comprehensively better compared to the state-of-the-art methods.

An Enhanced Latent Semantic Analysis Approach for Arabic Document Summarization

Purpose In general, cloud computing is a model of on-demand business computing that grants a convenient access to shared configurable resources on the internet. With the increment of workload and difficulty of tasks that are submitted by cloud consumers; “how to complete these tasks effectively and rapidly with limited cloud resources?” is becoming a challenging question. The major point of a task scheduling approach is to identify a trade-off among user needs and resource utilization. However, tasks that are submitted by varied users might have diverse needs of computing time, memory space, data traffic, response time, etc. This paper aims to proposes a new way of task scheduling. Design/methodology/approach To make the workflow completion in an efficient way and to reduce the cost and flow time, this paper proposes a new way of task scheduling. Here, a self-adaptive fruit fly optimization algorithm (SA-FFOA) is used for scheduling the workflow. The proposed multiple workflow scheduling model compares its efficiency over conventional methods in terms of analysis such as performance analysis, convergence analysis and statistical analysis. From the outcome of the analysis, the betterment of the proposed approach is proven with effective workflow scheduling. Findings The proposed algorithm is more superior regarding flow time with the minimum value, and the proposed model is enhanced over FFOA by 0.23%, differential evolution by 2.48%, artificial bee colony (ABC) by 2.85%, particle swarm optimization (PSO) by 2.46%, genetic algorithm (GA) by 2.33% and expected time to compute (ETC) by 2.56%. While analyzing the make span case, the proposed algorithm is 0.28%, 0.15%, 0.38%, 0.20%, 0.21% and 0.29% better than the conventional methods such as FFOA, DE, ABC, PSO, GA and ETC, respectively. Moreover, the proposed model has attained less cost, which is 2.14% better than FFOA, 2.32% better than DE, 3.53% better than ABC, 2.43% better than PSO, 2.07% better than GA and 2.90% better than ETC, respectively. Originality/value This paper presents a new way of task scheduling for making the workflow completion in an efficient way and for reducing the cost and flow time. This is the first paper uses SA-FFOA for scheduling the workflow.

Self adaptive fruit fly algorithm for multiple workflow scheduling in cloud computing environment

Study on Combinatorial Auction Mechanism for Resource Allocation in Cloud Computing Environment

On complex tasks scheduling scheme in cloud market based on coalition formation

http://or.nsfc.gov.cn/bitstream/00001903-5/329192/1/1000004221358.pdf

Ontology-based automatic summarization of web document

Multi-agent System is a hot topic of Artificial Intelligence, and it is extensively used to complete some tasks among different agents. While voting is often used for this purpose because it aggregates individual preferences into a collective decision. However, the winner determination problem has seriously hindered the development of the voting theory, then we innovatively introduce the concept of “satisfaction degree” to solve the problem. In this paper, we propose a formula for agents to express satisfaction degree of candidates, which we call Social Satisfaction Degree (SSD). To find the winners from candidates, we then design Single-winner Determination Algorithm (SWDA) and Multi-winner Determination Algorithm (MWDA) for single-winner and multi-winner based on SSD, respectively. The empirical results from the PrefLib data set show that our new algorithms can produce the winner set with optimal SSD more accurately than other voting rules.

A Voting Aggregation Algorithm for Optimal Social Satisfaction

As technology evolves and electricity demand rises, an increasing number of researches have focused on the efficient electricity allocation mechanisms so as to make consumer demand adaptive to the supply of electricity at all times. Considering the respective characteristics of the fixed price and the dynamic price model commonly used in the electricity market, in this paper, we formulate the problem of electricity allocation as a novel combinatorial auction model, and then put forward two directly applicable mechanisms called TAMEA-FP based on fixed price model and TAMEA-DP based on dynamic price model respectively. In addition, it is theoretically proven that both the proposed mechanisms are equipped with the economic properties such as individual rationality, budget balance and truthfulness. Finally, extensive simulation results show that both the proposed mechanisms own the allocation efficiency and computational traceability with the time complexity.

On truthful auction mechanisms for electricity allocation with multiple time slots

The invention provides an electric consumption transfer algorithm based on a continuous price function. Peak period and low period time segments are published by an intelligent electric network, two electric agents are designated according to geographic areas; the corresponding electric agents are selected by users according to the located geographic areas, and electric transfer cost and transfer capacity are submitted; winner sets are determined by the electric agents, after calculation, the transfer cost and the transfer capacity are submitted to the intelligent electric network; related parameters of a low period price function L(r) are adjusted by the intelligent electric network to satisfy balance distribution conditions; electric transfer amount and rewards are distributed by the intelligent electric network to the electric agents; the electric transfer amount and the rewards are distributed by the electric agents to the users; the distributed electric transfer amount is transferred by the users, and the users acquire the rewards.

Junwu Zhu

Papers

On complex tasks scheduling scheme in cloud market based on coalition formation

Ontology-based automatic summarization of web document

A Voting Aggregation Algorithm for Optimal Social Satisfaction

On truthful auction mechanisms for electricity allocation with multiple time slots

Electric consumption transfer algorithm based on continuous price function